ML -- I got about 2/3 of the way through that NFL book last night...some thoughts (spoils)
Posted by
Will Hunting (aka JoeMetz)
May 3 '12, 09:06
|
a) I think it's written poorly...redundant doesn't make the point stronger.
b) I think what very good points he brings up is tarnished a bit by some logical gaps in his process.
Notably, he spends a lot of time on taking apart people who say "Stats don't matter". Sure, over the long run (far more than 1 season of 16 games) stats matter, but given the idea of a football team is based on 1yr increments of 16 games....that's too small in my mind to destroy guys who say "Stats don't matter, wins do." He fails to articulate that if you flip the coin 16 times, and lose 75% of the flips, you get fired...but if you flip it 1600 times, the odds of you losing 75% go way down.
Becuase in 16 games, you can make the playoffs via a bad running game, or a bad passing game, or a weak defense..all you have to be is better than the opponent 9 times and it's likely you're in the mix.
Now, league wide...every season, you can get a picture of what stat trends = winning, but he's destroying individuals who see 16 results that matter before the fundamentals of a team can change.
He castigates people who deal in small sample sizes for relying on results and no real method to accrue more samples other than earning more seasons.
Other than that 10% of the book, I think he's got some ideas....but I don't think it's going to change the culture.
|
|
|